Palantir Technologies has collaborated with Nvidia to introduce a sovereign AI operating system reference architecture, aiming to streamline the deployment of AI data centres.

The new Palantir AI OS Reference Architecture (AIOS-RA) supports end-to-end processes from hardware acquisition to application rollout.

It operates on Nvidia’s Blackwell Ultra systems, which incorporate eight Blackwell Ultra GPUs and Spectrum-X Ethernet networking, supporting AI training and inference tasks.

The reference architecture integrates Palantir’s complete software suite, which includes AIP, Foundry, Apollo, Rubix, and AIP Hub.

It features a hardened Kubernetes substrate that runs various Foundry services such as Catalog, Build, and Multipass.

Management is handled through Palantir Rubix for zero-trust Kubernetes security and Apollo for autonomous deployment and lifecycle oversight.

Nvidia’s software stack underpins the infrastructure with components such as Nvidia AI Enterprise, CUDA-X Libraries, Nemotron open models, and Magnum IO for accelerated performance.

The architecture is designed to address requirements in environments with existing GPU resources, latency-sensitive operations, data sovereignty concerns, and broad geographic distribution.

Palantir chief architect Akshay Krishnaswamy said: “From our first deployment with the US government and in every deployment since, our software has had to meet the moment in the most complex and sensitive environments where customers must maintain control.

“Together with Nvidia — and building on many customers’ existing investments — we are proud to deliver a fully integrated AI operating system that is optimised for Nvidia accelerated compute infrastructure and enables customers to realise the promise of on-premise, edge, and sovereign cloud deployments.”

This collaboration grants enterprises full control over their data, AI models, and applications while supporting the use of open-source AI models and related data acceleration tools.

The unified system seeks to facilitate organisations in managing critical workloads on proprietary infrastructure.

Nvidia Enterprise AI Platforms vice president Justin Boitano said: “AI is redefining the infrastructure stack — demanding, latency-sensitive and data-sovereign environments require a full-stack architecture — built from silicon to systems to software.

“By combining Palantir’s sovereign AI OS reference architecture with Nvidia AI infrastructure, industries and nations can turn data into intelligence with speed, efficiency, and trust.”